Automated analysis of aerial photographs and potential for historic forest mapping
Abstract:Historic information regarding forest status is essential to management and conservation. Manual interpretation of aerial photography has long been the standard for forest inventory; however, manual interpretation can be subjective, inconsistent, and labor-intensive. This research compares automated techniques with manual interpretation results. First, we used an automated process (called segmentation) to delineate homogeneous stands of forests (or objects), analogous to the goal of manually delineating of polygons. Second, we used classification and regression tree (CART) analysis to classify polygons into the forest and terrain schemes used in British Columbia. Most characteristics of objects created via segmentation were similar to manually delineated polygons, as >70% of attributes were statistically similar across local, polygon, and landscape-level comparisons. Using manual interpretations for comparative reference, automated classifications produced overall accuracies ranging from 62% to 86% with per-class accuracies ranging from 0% to 96%. Automated methods yielded classifications meeting provincial overlap accuracy targets and helped identify classifications most suited to automation. Automated procedures have potential for aiding swift utilization of extensive historical photography archives with several caveats for future consideration. While automated techniques may never replicate all aspects of forest inventory classification, automated techniques may be valuable in assisting different phases of the process.
Document Type: Research Article
Affiliations: 1: Centre for Applied Conservation Research, Department of Forest and Conservation Sciences, 2424 Main Mall, University of British Columbia, Vancouver, BC, V6T 1Z4, Canada. 2: Centre for Applied Conservation Research, Department of Forest and Conservation Sciences, 2424 Main Mall, University of British Columbia, Vancouver, BC V6T 1Z4, Canada.
Publication date: 2013-05-21
- Published since 1971, this monthly journal features articles, reviews, notes and commentaries on all aspects of forest science, including biometrics and mensuration, conservation, disturbance, ecology, economics, entomology, fire, genetics, management, operations, pathology, physiology, policy, remote sensing, social science, soil, silviculture, wildlife and wood science, contributed by internationally respected scientists. It also publishes special issues dedicated to a topic of current interest.
- Information for Authors
- Submit a Paper
- Subscribe to this Title
- Terms & Conditions
- Sample Issue
- Reprints & Permissions
- Ingenta Connect is not responsible for the content or availability of external websites